Air Date: Week of June 9, 2017

The pruning process of evolution favors some patterns and disfavors others, says Dennett, and this has enabled life to evolve from single-celled bacteria (somewhat like the dividing cells above) to the complex array of life on Earth today. (Photo: NIAID, Flickr CC BY 2.0)

How machines learn is a growing branch of consciousness studies and one of the subjects addressed by analytic philosopher Daniel Dennett in his new book, “From Bacteria to Bach and Back: The Evolution of Minds.” Host Steve Curwood met Daniel Dennett at his Tufts University office where they discussed the origin of life, the evolution of language and why we should worry about our increasing dependence on the machines that have made modern life so convenient and simple.

Transcript

CURWOOD: It’s Living on Earth, I’m Steve Curwood. The analytic philosopher Daniel Dennett almost always seems to be thinking about the nature of consciousness, unless he’s writing about it, as in his newest book, “From Bacteria to Bach and Back”. The Tufts University professor has an enormous, white beard and a twinkle behind his spectacles, that makes him, well, rather look like Santa Claus, that is, if Santa Claus were a deep thinker who’d written a dozen books and spoke with a gently authoritative voice inflected by playful curiosity.

Professor Dennett looks to nature itself to help explain consciousness in a seemingly unconscious universe, peering through lenses that range from evolutionary biology to cognitive science and at analogs to nature such as artificial intelligence and computer learning. Daniel Dennett co-directs the Center for Cognitive Studies at Tufts University, where we spoke in his office. From the start it was obvious that even the simplest of questions, such as the usual public radio query that we pose to set recording levels for interviewees, “What did you have for breakfast?” could launch us into profound philosophical debate.

CURWOOD: Yes so what did you have for breakfast?

DENNETT: I had scrambled eggs this morning.

CURWOOD: Yeah, and where did the eggs come from?

DENNETT: Er, hens.

CURWOOD: And where did the hens come from?

DENNETT: Eggs.

CURWOOD: And the eggs came from?

DENNETT: Hens and so forth, but not ad infinitum.

CURWOOD: Oh, all right. So, where does it begin?

DENNETT: It begins with pre-genetic evolution of molecules. It begins with great whirling cycles of pre-biotic phenomena, and when you have hundreds of cycles at different revolutionary speeds, if you like, things gradually sort themselves out, just the way the sand and the pebbles do on a beach when the waves are coming in. It isn't just confetti, there's a pattern, and, as the pattern grows, some patterns are favored and other patterns are disfavored, and after a while, something happens which permits a different kind of periodicity which is, in the beginning, it's hard to say that it is reproduction when we think about what the first living thing or first reproducing thing was.

We tend to think about of some really simple bacteria, and I think that's almost certainly wrong. I think the first reproducing thing may have taken a thousand years to reproduce, but it reproduced. And gradually those early klugey Rube Goldberg reproducers got stream-lined. They had to because they were competing with each other. So, the stripped down, reproducing entity, that wasn't the origin of life. The origin of life was much earlier.

CURWOOD: Oh, OK. So, I know this is tough for a university professor, but give me the 30 second version of how, in your view, we evolved from microbes to the mind, or as you say your book, from bacteria to Bach.

DENNETT: Two processes had to happen. The first we all know about. It's Darwinian genetic evolution which created reptiles and then therapsids and then mammals and then primates along with all the trees and flowers and fish and all the rest of it. And something not miraculous, but something pretty unusual, happened, and there's a great debate about that, but, whatever happened, it gave birth to language, and language changed everything because language, like DNA, is digitized. We have phonemes which are like the ACGT of DNA, one of nature's great inventions. No person invented it, but once you have phonemes, you can transmit hi-fidelity information from place to place. Our ancestors from over thousands of years, maybe hundreds of thousands of years, they came to understand what it was, "Hey, we're talking", and it turns out that talking is just the most potent transmitter of information yet.

CURWOOD: Indeed, as we know on the radio right?

DENNETT: Absolutely. The great thing about talking is that you can learn things that your parents don't know. Genetically, we can inherit a lot of know-how, a lot of instincts, a lot of talents, the ability to digest our food, and to walk and all kinds of things are actually passed on through the genome pretty much the way a bird's ability to build a nest is passed on. They don't have to have seen a nest built to know how to build a nest. That goes on through the genes, but we, and we only, have this other information highway primarily made of language which makes it possible for us to learn from people that aren't related to us at all, from Socrates and Plato and to Galileo and from people living on the other side of the globe today, we have all of these sources of information, and what it does is, it multiplies by many orders of magnitude the ability of an individual organism to take on new competence and new comprehension.

CURWOOD: So, you say that there are two important figures that got you to this notion of evolutionary thinking. How did they change things?

DENNETT: Well, my two real heroes here are two Brits, Charles Darwin, and a hundred years later or so, Alan Turing. And their revolutions were actually similar. They're two faces of the same coin. What Darwin showed, as one of his early critics said, is that absolute ignorance can take the place of intelligence in the creation of all the wonders of the biosphere. The process of natural selection isn't the mind. It isn't intelligent. It's just a brute, mechanical algorithmic process that churns away mindlessly over the millions and millions of years, billions of years, and creates the bird's wing and the legs of the cheetah and all of the rest of the wonders of the natural world.

Turing had a very similar message. In his day, a computer was a person, typically a woman, who sat and computed, and he realized that the jobs they were doing didn’t require comprehension. You didn't have to have a degree in mathematics to do them. You could be entirely mechanized, and so he created the idea of an automatic mechanical mindless computer, and he described their smallest parts, bits, binary digits, zero or one, and a few basic operations, plus and minus and compared to zero, and you can get what's called conditional branching, and you're off to the races, and you get compute anything computable. So, again, what he was showing was you can have competence without comprehension, and he went on to say what comprehension is, in effect, is just layers and layers and layers and layers of competence that build on each other.

And here's where the Darwinian idea of gradualism comes in. There's no moment when a bell rings, and -- ta-da! -- we have comprehension. The model for comprehension is spending 12 years in school and coming out, not just competent, but being able to explain what the competence consists in, not perfectly but pretty well.

CURWOOD: Let's talk about competence without comprehension. In your book, you write about an interesting example, and that's the termite mound, and, you know, it looks remarkably like big castles, a place in Barcelona you point out that it could look like. Why this is comparison important, do you think?

DENNETT: Well, here we have two artifacts, the famous church of church of Gaudi, La Sagrada Familia in Barcelona, and the termite castle, and they look almost identical and, in fact, there's many structural details in common too. So, here we have two artifacts made by living things. They profoundly different in the research and development and building that went into them.

The termite castle is built by clueless termites, there's no architect, there's no leader. They just go out there and they do their little mindless thing and lo, and behold, this amazing structure appears. That's Darwinian research and development, competence without comprehension.

Gaudi's church, on the other hand, is a brilliant example of a charismatic genius autocratic work. He writes manifestos. He has the blueprints. He's ordering his subordinates around who order their subordinates around. This is top-down intelligent design, xmall "i", small "d". So, if we think of Gaudi as a sort of model of an intelligent designer and the termites as a model of an uncomprehending builder, then we have a profound gap between them. So, the question is, how do you get to Gaudi from the termite? How do you get to Bach from the bacterium?

CURWOOD: Yes, tell me.

DENNETT: And what particularly puts a point on this is, if you think about our brains, what we have between our ears is by latest count about 86 billion neurons, and each one of those is clueless. I mean, two million termites can make a castle, but they can't do it Gaudi's way. But you put 86 billion neurons in a brain, and, under the right conditions, you get science. You get art. You get politics. You get human culture, and the answer, I think, is and has to be by a second evolutionary process just as non-miraculous as the first, and that's cultural evolution. Once we have language, it starts out with our first speakers about as clueless as chimpanzees, let's say. Chimpanzees have maybe a dozen things that they can pass on to their young without going through the genes. We have hundreds and hundreds of thousands of things that we can pass on because our culture is cumulative.

CURWOOD: So, you're saying that language itself evolved on a track that is parallel to life itself by natural selection. How does that work?

DENNETT: Let's take words. Words have lineages. The word “table” comes from the Latin word “tabula”, which has earlier roots. So, if you think of a word as not alive but as subject to natural selection, words are competing for space in people's brains, and words go extinct. Words mutate. They change their meaning, their pronunciation.

Now, Darwin noticed this, and Richard Dawkins made a big thing of this in his book "The Selfish Gene", and he coined the term “meme”, which is more general than just words. There are ways of doing things like wearing your baseball cap backwards or dancing the foxtrot or kneading your bread just so and making an arch. Those things can be passed on non-genetically, and they evolve, accumulating mutations as they go, but you have to recognize that words and other memes, they have their own fitness, and that's key because we want to be able to explain all of human culture, not just the good stuff. There's junk and ugly practices, just bad habits. The point is, bad habits can evolve just as good habits do. Why? Because they're just more infectious. They grab onto some weakness in our makeup, in our dispositions, and they exploit it, just like viruses.

CURWOOD: Let's talk about the human mind and the machines, the computers, and what do you see is the next step in machine learning?

DENNETT: Modern day human beings are wonderfully intelligent designers, engineers, scientists, poets, artists, composers, but one thing we've learned, thanks to Darwin and Turing and others, is that evolution is cleverer than we are. So, now we're learning how to harness evolutionary and quasi-evolutionary processes to do the heavy lifting for us. So, we're creating a world in which we have these new oracle boxes that give us answers to our questions. Unlike the oracles of ancient times, we have very, very good reasons to believe what they tell us most of the time. But we can't understand exactly how we get their results, and that's a bit scary.

CURWOOD: So, you're talking about the risks of becoming so dependent on our technology.

DENNETT: Absolutely. We have becoming breathtakingly dependent on technology that only a few of us know how to understand or repair. It used to be that, if your car broke down and you were a clever person, and had taken a few courses in automotive mechanics, you could fix your car. That day is disappearing fast. Now, of course, we're also losing our talent to read maps because GPS is taking over, and the fact is that if we were to have an electronic blackout, we would be thrust into the most destructive panic you can imagine. Of course, it would mean we would be suddenly thrust back into the 19th century. Well, life was pretty good in the 19th century. You could stay alive and comfortable and read books and have food and shelter and all the rest. We could in principle go back to that, but only if we anticipate that we might have to and think about who is going to know how to shoe a horse and who's going know how to make a steam engine that you fire up with firewood or something like that. Some people pride themselves in their sort of practical knowledge of the 19th century ways. They might be very important to us if we ever have that technological crash, which I think is distinctly possible.

CURWOOD: Some people worry that, as computers advance, as artificial intelligence advances, that the machines will take over. How much of that concern do you share?

DENNETT: Well, I think it's already happened in some regards. I don't think we have to worry about, you know, robots enslaving us, but I think we have to worry about our own dependence on them. How much of our annual income goes to feeding the beast, buying the upgrades, keeping our cell phones and computers, which nobody needed before, now everybody needs one?

I once gave my students a fantasy about this gift from outer space that arrives, and scientists study it and sort of figure out, reverse engineer it, and then they start make thousands and thousands and millions of copies, and everyone wants to have one. And pretty soon everybody on the planet is working away so that they can have these marvelous magic boxes. We've become enslaved by these things. The only thing false about that is, it came from outer space. We got them. I think strong artificial intelligence, agents that are conscious and that treat us like dirt or slaves or whatever, that's a fantasy that's not just unrealistic, I think it's harmful because it gets us all worried about something which is hundreds of years away instead of worrying about the thing which is right here now.

CURWOOD: Daniel Dennett is a philosopher and he co-directs the Center for Cognitive Studies at Tufts University. His new book is called "From Bacteria to Bach and Back: The evolution of Minds". Thanks so much for taking the time today.

DENNETT: Thanks so much. I enjoyed the interview.

 

Links

From Bacteria to Bach and Back

About Daniel Dennett